agora inbox for [email protected]help / color / mirror / Atom feed
[PATCH v28 10/11] Add regression tests for Incremental View Maintenance 426+ messages / 3 participants [nested] [flat]
* [PATCH v28 10/11] Add regression tests for Incremental View Maintenance @ 2021-03-10 02:11 Takuma Hoshiai <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Takuma Hoshiai @ 2021-03-10 02:11 UTC (permalink / raw) --- src/bin/pg_dump/pg_dump.h | 1 + .../regress/expected/incremental_matview.out | 1031 +++++++++++++++++ src/test/regress/parallel_schedule | 2 +- src/test/regress/sql/incremental_matview.sql | 534 +++++++++ 4 files changed, 1567 insertions(+), 1 deletion(-) create mode 100644 src/test/regress/expected/incremental_matview.out create mode 100644 src/test/regress/sql/incremental_matview.sql diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h index 119293a751..c3960a4192 100644 --- a/src/bin/pg_dump/pg_dump.h +++ b/src/bin/pg_dump/pg_dump.h @@ -322,6 +322,7 @@ typedef struct _tableInfo int numParents; /* number of (immediate) parent tables */ struct _tableInfo **parents; /* TableInfos of immediate parents */ + bool isivm; /* is incrementally maintainable materialized view? */ /* diff --git a/src/test/regress/expected/incremental_matview.out b/src/test/regress/expected/incremental_matview.out new file mode 100644 index 0000000000..b1e9424508 --- /dev/null +++ b/src/test/regress/expected/incremental_matview.out @@ -0,0 +1,1031 @@ +-- create a table to use as a basis for views and materialized views in various combinations +CREATE TABLE mv_base_a (i int, j int); +INSERT INTO mv_base_a VALUES + (1,10), + (2,20), + (3,30), + (4,40), + (5,50); +CREATE TABLE mv_base_b (i int, k int); +INSERT INTO mv_base_b VALUES + (1,101), + (2,102), + (3,103), + (4,104); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_1 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) WITH NO DATA; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; +ERROR: materialized view "mv_ivm_1" has not been populated +HINT: Use the REFRESH MATERIALIZED VIEW command. +REFRESH MATERIALIZED VIEW mv_ivm_1; +NOTICE: could not create an index on materialized view "mv_ivm_1" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + i | j | k +---+----+----- + 1 | 10 | 101 + 2 | 20 | 102 + 3 | 30 | 103 + 4 | 40 | 104 +(4 rows) + +-- REFRESH WITH NO DATA +BEGIN; +CREATE FUNCTION dummy_ivm_trigger_func() RETURNS TRIGGER AS $$ + BEGIN + RETURN NULL; + END +$$ language plpgsql; +CREATE CONSTRAINT TRIGGER dummy_ivm_trigger AFTER INSERT +ON mv_base_a FROM mv_ivm_1 FOR EACH ROW +EXECUTE PROCEDURE dummy_ivm_trigger_func(); +SELECT COUNT(*) +FROM pg_depend pd INNER JOIN pg_trigger pt ON pd.objid = pt.oid +WHERE pd.classid = 'pg_trigger'::regclass AND pd.refobjid = 'mv_ivm_1'::regclass; + count +------- + 17 +(1 row) + +REFRESH MATERIALIZED VIEW mv_ivm_1 WITH NO DATA; +SELECT COUNT(*) +FROM pg_depend pd INNER JOIN pg_trigger pt ON pd.objid = pt.oid +WHERE pd.classid = 'pg_trigger'::regclass AND pd.refobjid = 'mv_ivm_1'::regclass; + count +------- + 1 +(1 row) + +ROLLBACK; +-- immediate maintenance +BEGIN; +INSERT INTO mv_base_b VALUES(5,105); +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + i | j | k +---+----+----- + 1 | 10 | 101 + 2 | 20 | 102 + 3 | 30 | 103 + 4 | 40 | 104 + 5 | 50 | 105 +(5 rows) + +UPDATE mv_base_a SET j = 0 WHERE i = 1; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + i | j | k +---+----+----- + 1 | 0 | 101 + 2 | 20 | 102 + 3 | 30 | 103 + 4 | 40 | 104 + 5 | 50 | 105 +(5 rows) + +DELETE FROM mv_base_b WHERE (i,k) = (5,105); +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + i | j | k +---+----+----- + 1 | 0 | 101 + 2 | 20 | 102 + 3 | 30 | 103 + 4 | 40 | 104 +(4 rows) + +ROLLBACK; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + i | j | k +---+----+----- + 1 | 10 | 101 + 2 | 20 | 102 + 3 | 30 | 103 + 4 | 40 | 104 +(4 rows) + +-- rename of IVM columns +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_rename AS SELECT DISTINCT * FROM mv_base_a; +NOTICE: created index "mv_ivm_rename_index" on materialized view "mv_ivm_rename" +ALTER MATERIALIZED VIEW mv_ivm_rename RENAME COLUMN __ivm_count__ TO xxx; +ERROR: IVM column can not be renamed +DROP MATERIALIZED VIEW mv_ivm_rename; +-- unique index on IVM columns +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_unique AS SELECT DISTINCT * FROM mv_base_a; +NOTICE: created index "mv_ivm_unique_index" on materialized view "mv_ivm_unique" +CREATE UNIQUE INDEX ON mv_ivm_unique(__ivm_count__); +ERROR: unique index creation on IVM columns is not supported +CREATE UNIQUE INDEX ON mv_ivm_unique((__ivm_count__)); +ERROR: unique index creation on IVM columns is not supported +CREATE UNIQUE INDEX ON mv_ivm_unique((__ivm_count__ + 1)); +ERROR: unique index creation on IVM columns is not supported +DROP MATERIALIZED VIEW mv_ivm_unique; +-- TRUNCATE a base table in join views +BEGIN; +TRUNCATE mv_base_a; +SELECT * FROM mv_ivm_1; + i | j | k +---+---+--- +(0 rows) + +ROLLBACK; +BEGIN; +TRUNCATE mv_base_b; +SELECT * FROM mv_ivm_1; + i | j | k +---+---+--- +(0 rows) + +ROLLBACK; +-- some query syntax +BEGIN; +CREATE FUNCTION ivm_func() RETURNS int LANGUAGE 'sql' + AS 'SELECT 1' IMMUTABLE; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_func AS SELECT * FROM ivm_func(); +NOTICE: could not create an index on materialized view "mv_ivm_func" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_no_tbl AS SELECT 1; +NOTICE: could not create an index on materialized view "mv_ivm_no_tbl" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +ROLLBACK; +-- result of materialized view have DISTINCT clause or the duplicate result. +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_duplicate AS SELECT j FROM mv_base_a; +NOTICE: could not create an index on materialized view "mv_ivm_duplicate" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_distinct AS SELECT DISTINCT j FROM mv_base_a; +NOTICE: created index "mv_ivm_distinct_index" on materialized view "mv_ivm_distinct" +INSERT INTO mv_base_a VALUES(6,20); +SELECT * FROM mv_ivm_duplicate ORDER BY 1; + j +---- + 10 + 20 + 20 + 30 + 40 + 50 +(6 rows) + +SELECT * FROM mv_ivm_distinct ORDER BY 1; + j +---- + 10 + 20 + 30 + 40 + 50 +(5 rows) + +DELETE FROM mv_base_a WHERE (i,j) = (2,20); +SELECT * FROM mv_ivm_duplicate ORDER BY 1; + j +---- + 10 + 20 + 30 + 40 + 50 +(5 rows) + +SELECT * FROM mv_ivm_distinct ORDER BY 1; + j +---- + 10 + 20 + 30 + 40 + 50 +(5 rows) + +ROLLBACK; +-- support SUM(), COUNT() and AVG() aggregate functions +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(i), AVG(j) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_agg_index" on materialized view "mv_ivm_agg" +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 20 | 1 | 20.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +INSERT INTO mv_base_a VALUES(2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 120 | 2 | 60.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +UPDATE mv_base_a SET j = 200 WHERE (i,j) = (2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; + i | sum | count | avg +---+-----+-------+---------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 220 | 2 | 110.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +DELETE FROM mv_base_a WHERE (i,j) = (2,200); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 20 | 1 | 20.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +ROLLBACK; +-- support COUNT(*) aggregate function +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_agg_index" on materialized view "mv_ivm_agg" +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3; + i | sum | count +---+-----+------- + 1 | 10 | 1 + 2 | 20 | 1 + 3 | 30 | 1 + 4 | 40 | 1 + 5 | 50 | 1 +(5 rows) + +INSERT INTO mv_base_a VALUES(2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3; + i | sum | count +---+-----+------- + 1 | 10 | 1 + 2 | 120 | 2 + 3 | 30 | 1 + 4 | 40 | 1 + 5 | 50 | 1 +(5 rows) + +ROLLBACK; +-- TRUNCATE a base table in aggregate views +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_agg_index" on materialized view "mv_ivm_agg" +TRUNCATE mv_base_a; +SELECT sum, count FROM mv_ivm_agg; + sum | count +-----+------- +(0 rows) + +SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; + i | sum | count +---+-----+------- +(0 rows) + +ROLLBACK; +-- support aggregate functions without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_group AS SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; +NOTICE: could not create an index on materialized view "mv_ivm_group" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv_ivm_group ORDER BY 1; + sum | count | avg +-----+-------+--------------------- + 150 | 5 | 30.0000000000000000 +(1 row) + +INSERT INTO mv_base_a VALUES(6,60); +SELECT * FROM mv_ivm_group ORDER BY 1; + sum | count | avg +-----+-------+--------------------- + 210 | 6 | 35.0000000000000000 +(1 row) + +DELETE FROM mv_base_a; +SELECT * FROM mv_ivm_group ORDER BY 1; + sum | count | avg +-----+-------+----- + | 0 | +(1 row) + +ROLLBACK; +-- TRUNCATE a base table in aggregate views without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_group AS SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; +NOTICE: could not create an index on materialized view "mv_ivm_group" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +TRUNCATE mv_base_a; +SELECT sum, count, avg FROM mv_ivm_group; + sum | count | avg +-----+-------+----- + | 0 | +(1 row) + +SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; + sum | count | avg +-----+-------+----- + | 0 | +(1 row) + +ROLLBACK; +-- resolved issue: When use AVG() function and values is indivisible, result of AVG() is incorrect. +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_avg_bug AS SELECT i, SUM(j), COUNT(j), AVG(j) FROM mv_base_A GROUP BY i; +NOTICE: created index "mv_ivm_avg_bug_index" on materialized view "mv_ivm_avg_bug" +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 20 | 1 | 20.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +INSERT INTO mv_base_a VALUES + (1,0), + (1,0), + (2,30), + (2,30); +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 3 | 3.3333333333333333 + 2 | 80 | 3 | 26.6666666666666667 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +DELETE FROM mv_base_a WHERE (i,j) = (1,0); +DELETE FROM mv_base_a WHERE (i,j) = (2,30); +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; + i | sum | count | avg +---+-----+-------+--------------------- + 1 | 10 | 1 | 10.0000000000000000 + 2 | 20 | 1 | 20.0000000000000000 + 3 | 30 | 1 | 30.0000000000000000 + 4 | 40 | 1 | 40.0000000000000000 + 5 | 50 | 1 | 50.0000000000000000 +(5 rows) + +ROLLBACK; +-- support MIN(), MAX() aggregate functions +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min_max AS SELECT i, MIN(j), MAX(j) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_min_max_index" on materialized view "mv_ivm_min_max" +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; + i | min | max +---+-----+----- + 1 | 10 | 10 + 2 | 20 | 20 + 3 | 30 | 30 + 4 | 40 | 40 + 5 | 50 | 50 +(5 rows) + +INSERT INTO mv_base_a VALUES + (1,11), (1,12), + (2,21), (2,22), + (3,31), (3,32), + (4,41), (4,42), + (5,51), (5,52); +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; + i | min | max +---+-----+----- + 1 | 10 | 12 + 2 | 20 | 22 + 3 | 30 | 32 + 4 | 40 | 42 + 5 | 50 | 52 +(5 rows) + +DELETE FROM mv_base_a WHERE (i,j) IN ((1,10), (2,21), (3,32)); +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; + i | min | max +---+-----+----- + 1 | 11 | 12 + 2 | 20 | 22 + 3 | 30 | 31 + 4 | 40 | 42 + 5 | 50 | 52 +(5 rows) + +ROLLBACK; +-- support MIN(), MAX() aggregate functions without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min_max AS SELECT MIN(j), MAX(j) FROM mv_base_a; +NOTICE: could not create an index on materialized view "mv_ivm_min_max" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv_ivm_min_max; + min | max +-----+----- + 10 | 50 +(1 row) + +INSERT INTO mv_base_a VALUES + (0,0), (6,60), (7,70); +SELECT * FROM mv_ivm_min_max; + min | max +-----+----- + 0 | 70 +(1 row) + +DELETE FROM mv_base_a WHERE (i,j) IN ((0,0), (7,70)); +SELECT * FROM mv_ivm_min_max; + min | max +-----+----- + 10 | 60 +(1 row) + +DELETE FROM mv_base_a; +SELECT * FROM mv_ivm_min_max; + min | max +-----+----- + | +(1 row) + +ROLLBACK; +-- Test MIN/MAX after search_path change +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min AS SELECT MIN(j) FROM mv_base_a; +NOTICE: could not create an index on materialized view "mv_ivm_min" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv_ivm_min; + min +----- + 10 +(1 row) + +CREATE SCHEMA myschema; +GRANT ALL ON SCHEMA myschema TO public; +CREATE TABLE myschema.mv_base_a (j int); +INSERT INTO myschema.mv_base_a VALUES (1); +DELETE FROM mv_base_a WHERE (i,j) = (1,10); +SELECT * FROM mv_ivm_min; + min +----- + 20 +(1 row) + +SET search_path TO myschema,public,pg_catalog; +DELETE FROM public.mv_base_a WHERE (i,j) = (2,20); +SELECT * FROM mv_ivm_min; + min +----- + 30 +(1 row) + +ROLLBACK; +-- aggregate views with column names specified +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_agg_index" on materialized view "mv_ivm_agg" +INSERT INTO mv_base_a VALUES (1,100), (2,200), (3,300); +UPDATE mv_base_a SET j = 2000 WHERE (i,j) = (2,20); +DELETE FROM mv_base_a WHERE (i,j) = (3,30); +SELECT * FROM mv_ivm_agg ORDER BY 1,2; + a | sum +---+------ + 1 | 110 + 2 | 2200 + 3 | 300 + 4 | 40 + 5 | 50 +(5 rows) + +ROLLBACK; +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a,b) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +NOTICE: created index "mv_ivm_agg_index" on materialized view "mv_ivm_agg" +INSERT INTO mv_base_a VALUES (1,100), (2,200), (3,300); +UPDATE mv_base_a SET j = 2000 WHERE (i,j) = (2,20); +DELETE FROM mv_base_a WHERE (i,j) = (3,30); +SELECT * FROM mv_ivm_agg ORDER BY 1,2; + a | b +---+------ + 1 | 110 + 2 | 2200 + 3 | 300 + 4 | 40 + 5 | 50 +(5 rows) + +ROLLBACK; +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a,b,c) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +ERROR: too many column names were specified +ROLLBACK; +-- support self join view and multiple change on the same table +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (1, 10), (2, 20), (3, 30); +CREATE INCREMENTAL MATERIALIZED VIEW mv_self(v1, v2) AS + SELECT t1.v, t2.v FROM base_t AS t1 JOIN base_t AS t2 ON t1.i = t2.i; +NOTICE: could not create an index on materialized view "mv_self" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv_self ORDER BY v1; + v1 | v2 +----+---- + 10 | 10 + 20 | 20 + 30 | 30 +(3 rows) + +INSERT INTO base_t VALUES (4,40); +DELETE FROM base_t WHERE i = 1; +UPDATE base_t SET v = v*10 WHERE i=2; +SELECT * FROM mv_self ORDER BY v1; + v1 | v2 +-----+----- + 30 | 30 + 40 | 40 + 200 | 200 +(3 rows) + +WITH + ins_t1 AS (INSERT INTO base_t VALUES (5,50) RETURNING 1), + ins_t2 AS (INSERT INTO base_t VALUES (6,60) RETURNING 1), + upd_t AS (UPDATE base_t SET v = v + 100 RETURNING 1), + dlt_t AS (DELETE FROM base_t WHERE i IN (4,5) RETURNING 1) +SELECT NULL; + ?column? +---------- + +(1 row) + +SELECT * FROM mv_self ORDER BY v1; + v1 | v2 +-----+----- + 50 | 50 + 60 | 60 + 130 | 130 + 300 | 300 +(4 rows) + +--- with sub-transactions +SAVEPOINT p1; +INSERT INTO base_t VALUES (7,70); +RELEASE SAVEPOINT p1; +INSERT INTO base_t VALUES (7,77); +SELECT * FROM mv_self ORDER BY v1, v2; + v1 | v2 +-----+----- + 50 | 50 + 60 | 60 + 70 | 70 + 70 | 77 + 77 | 70 + 77 | 77 + 130 | 130 + 300 | 300 +(8 rows) + +ROLLBACK; +-- support simultaneous table changes +BEGIN; +CREATE TABLE base_r (i int, v int); +CREATE TABLE base_s (i int, v int); +INSERT INTO base_r VALUES (1, 10), (2, 20), (3, 30); +INSERT INTO base_s VALUES (1, 100), (2, 200), (3, 300); +CREATE INCREMENTAL MATERIALIZED VIEW mv(v1, v2) AS + SELECT r.v, s.v FROM base_r AS r JOIN base_s AS s USING(i); +NOTICE: could not create an index on materialized view "mv" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv ORDER BY v1; + v1 | v2 +----+----- + 10 | 100 + 20 | 200 + 30 | 300 +(3 rows) + +WITH + ins_r AS (INSERT INTO base_r VALUES (1,11) RETURNING 1), + ins_r2 AS (INSERT INTO base_r VALUES (3,33) RETURNING 1), + ins_s AS (INSERT INTO base_s VALUES (2,222) RETURNING 1), + upd_r AS (UPDATE base_r SET v = v + 1000 WHERE i = 2 RETURNING 1), + dlt_s AS (DELETE FROM base_s WHERE i = 3 RETURNING 1) +SELECT NULL; + ?column? +---------- + +(1 row) + +SELECT * FROM mv ORDER BY v1; + v1 | v2 +------+----- + 10 | 100 + 11 | 100 + 1020 | 200 + 1020 | 222 +(4 rows) + +ROLLBACK; +-- support foreign reference constraints +BEGIN; +CREATE TABLE ri1 (i int PRIMARY KEY); +CREATE TABLE ri2 (i int PRIMARY KEY REFERENCES ri1(i) ON UPDATE CASCADE ON DELETE CASCADE, v int); +INSERT INTO ri1 VALUES (1),(2),(3); +INSERT INTO ri2 VALUES (1),(2),(3); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ri(i1, i2) AS + SELECT ri1.i, ri2.i FROM ri1 JOIN ri2 USING(i); +NOTICE: created index "mv_ri_index" on materialized view "mv_ri" +SELECT * FROM mv_ri ORDER BY i1; + i1 | i2 +----+---- + 1 | 1 + 2 | 2 + 3 | 3 +(3 rows) + +UPDATE ri1 SET i=10 where i=1; +DELETE FROM ri1 WHERE i=2; +SELECT * FROM mv_ri ORDER BY i2; + i1 | i2 +----+---- + 3 | 3 + 10 | 10 +(2 rows) + +ROLLBACK; +-- views including NULL +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (1,10),(2, NULL); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT * FROM base_t; +NOTICE: could not create an index on materialized view "mv" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv ORDER BY i; + i | v +---+---- + 1 | 10 + 2 | +(2 rows) + +UPDATE base_t SET v = 20 WHERE i = 2; +SELECT * FROM mv ORDER BY i; + i | v +---+---- + 1 | 10 + 2 | 20 +(2 rows) + +ROLLBACK; +BEGIN; +CREATE TABLE base_t (i int); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT * FROM base_t; +NOTICE: could not create an index on materialized view "mv" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT * FROM mv ORDER BY i; + i +--- +(0 rows) + +INSERT INTO base_t VALUES (1),(NULL); +SELECT * FROM mv ORDER BY i; + i +--- + 1 + +(2 rows) + +ROLLBACK; +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (NULL, 1), (NULL, 2), (1, 10), (1, 20); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT i, sum(v) FROM base_t GROUP BY i; +NOTICE: created index "mv_index" on materialized view "mv" +SELECT * FROM mv ORDER BY i; + i | sum +---+----- + 1 | 30 + | 3 +(2 rows) + +UPDATE base_t SET v = v * 10; +SELECT * FROM mv ORDER BY i; + i | sum +---+----- + 1 | 300 + | 30 +(2 rows) + +ROLLBACK; +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (NULL, 1), (NULL, 2), (NULL, 3), (NULL, 4), (NULL, 5); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT i, min(v), max(v) FROM base_t GROUP BY i; +NOTICE: created index "mv_index" on materialized view "mv" +SELECT * FROM mv ORDER BY i; + i | min | max +---+-----+----- + | 1 | 5 +(1 row) + +DELETE FROM base_t WHERE v = 1; +SELECT * FROM mv ORDER BY i; + i | min | max +---+-----+----- + | 2 | 5 +(1 row) + +DELETE FROM base_t WHERE v = 3; +SELECT * FROM mv ORDER BY i; + i | min | max +---+-----+----- + | 2 | 5 +(1 row) + +DELETE FROM base_t WHERE v = 5; +SELECT * FROM mv ORDER BY i; + i | min | max +---+-----+----- + | 2 | 4 +(1 row) + +ROLLBACK; +-- IMMV containing user defined type +BEGIN; +CREATE TYPE mytype; +CREATE FUNCTION mytype_in(cstring) + RETURNS mytype AS 'int4in' + LANGUAGE INTERNAL STRICT IMMUTABLE; +NOTICE: return type mytype is only a shell +CREATE FUNCTION mytype_out(mytype) + RETURNS cstring AS 'int4out' + LANGUAGE INTERNAL STRICT IMMUTABLE; +NOTICE: argument type mytype is only a shell +CREATE TYPE mytype ( + LIKE = int4, + INPUT = mytype_in, + OUTPUT = mytype_out +); +CREATE FUNCTION mytype_eq(mytype, mytype) + RETURNS bool AS 'int4eq' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE FUNCTION mytype_lt(mytype, mytype) + RETURNS bool AS 'int4lt' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE FUNCTION mytype_cmp(mytype, mytype) + RETURNS integer AS 'btint4cmp' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE OPERATOR = ( + leftarg = mytype, rightarg = mytype, + procedure = mytype_eq); +CREATE OPERATOR < ( + leftarg = mytype, rightarg = mytype, + procedure = mytype_lt); +CREATE OPERATOR CLASS mytype_ops + DEFAULT FOR TYPE mytype USING btree AS + OPERATOR 1 <, + OPERATOR 3 = , + FUNCTION 1 mytype_cmp(mytype,mytype); +CREATE TABLE t_mytype (x mytype); +CREATE INCREMENTAL MATERIALIZED VIEW mv_mytype AS + SELECT * FROM t_mytype; +NOTICE: could not create an index on materialized view "mv_mytype" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +INSERT INTO t_mytype VALUES ('1'::mytype); +SELECT * FROM mv_mytype; + x +--- + 1 +(1 row) + +ROLLBACK; +-- outer join is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv(a,b) AS SELECT a.i, b.i FROM mv_base_a a LEFT JOIN mv_base_b b ON a.i=b.i; +ERROR: OUTER JOIN is not supported on incrementally maintainable materialized view +-- CTE is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv AS + WITH b AS ( SELECT * FROM mv_base_b) SELECT a.i,a.j FROM mv_base_a a, b WHERE a.i = b.i; +ERROR: CTE is not supported on incrementally maintainable materialized view +-- contain system column +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm01 AS SELECT i,j,xmin FROM mv_base_a; +ERROR: system column is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm02 AS SELECT i,j FROM mv_base_a WHERE xmin = '610'; +ERROR: system column is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm04 AS SELECT i,j,xmin::text AS x_min FROM mv_base_a; +ERROR: system column is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm06 AS SELECT i,j,xidsend(xmin) AS x_min FROM mv_base_a; +ERROR: system column is not supported on incrementally maintainable materialized view +-- contain subquery +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm03 AS SELECT i,j FROM mv_base_a WHERE i IN (SELECT i FROM mv_base_b WHERE k < 103 ); +ERROR: subquery is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm04 AS SELECT a.i,a.j FROM mv_base_a a, (SELECT * FROM mv_base_b) b WHERE a.i = b.i; +ERROR: subquery is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm05 AS SELECT i,j, (SELECT k FROM mv_base_b b WHERE a.i = b.i) FROM mv_base_a a; +ERROR: subquery is not supported on incrementally maintainable materialized view +-- contain ORDER BY +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm07 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) ORDER BY i,j,k; +ERROR: ORDER BY clause is not supported on incrementally maintainable materialized view +-- contain HAVING +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm08 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) GROUP BY i,j,k HAVING SUM(i) > 5; +ERROR: HAVING clause is not supported on incrementally maintainable materialized view +-- contain view or materialized view +CREATE VIEW b_view AS SELECT i,k FROM mv_base_b; +CREATE MATERIALIZED VIEW b_mview AS SELECT i,k FROM mv_base_b; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm07 AS SELECT a.i,a.j FROM mv_base_a a,b_view b WHERE a.i = b.i; +ERROR: VIEW or MATERIALIZED VIEW is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm08 AS SELECT a.i,a.j FROM mv_base_a a,b_mview b WHERE a.i = b.i; +ERROR: VIEW or MATERIALIZED VIEW is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm09 AS SELECT a.i,a.j FROM mv_base_a a, (SELECT i, COUNT(*) FROM mv_base_b GROUP BY i) b WHERE a.i = b.i; +ERROR: subquery is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm10 AS SELECT a.i,a.j FROM mv_base_a a WHERE EXISTS(SELECT 1 FROM mv_base_b b WHERE a.i = b.i) OR a.i > 5; +ERROR: subquery is not supported on incrementally maintainable materialized view +-- contain mutable functions +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm12 AS SELECT i,j FROM mv_base_a WHERE i = random()::int; +ERROR: mutable function is not supported on incrementally maintainable materialized view +HINT: functions must be marked IMMUTABLE +-- LIMIT/OFFSET is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm13 AS SELECT i,j FROM mv_base_a LIMIT 10 OFFSET 5; +ERROR: LIMIT/OFFSET clause is not supported on incrementally maintainable materialized view +-- DISTINCT ON is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm14 AS SELECT DISTINCT ON(i) i, j FROM mv_base_a; +ERROR: DISTINCT ON is not supported on incrementally maintainable materialized view +-- TABLESAMPLE clause is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm15 AS SELECT i, j FROM mv_base_a TABLESAMPLE SYSTEM(50); +ERROR: TABLESAMPLE clause is not supported on incrementally maintainable materialized view +-- window functions are not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm16 AS SELECT *, cume_dist() OVER (ORDER BY i) AS rank FROM mv_base_a; +ERROR: window functions are not supported on incrementally maintainable materialized view +-- aggregate function with some options is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm17 AS SELECT COUNT(*) FILTER(WHERE i < 3) FROM mv_base_a; +ERROR: aggregate function with FILTER clause is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm18 AS SELECT COUNT(DISTINCT i) FROM mv_base_a; +ERROR: aggregate function with DISTINCT arguments is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm19 AS SELECT array_agg(j ORDER BY i DESC) FROM mv_base_a; +ERROR: aggregate function with ORDER clause is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm20 AS SELECT i,SUM(j) FROM mv_base_a GROUP BY GROUPING SETS((i),()); +ERROR: GROUPING SETS, ROLLUP, or CUBE clauses is not supported on incrementally maintainable materialized view +-- inheritance parent is not supported +BEGIN; +CREATE TABLE parent (i int, v int); +CREATE TABLE child_a(options text) INHERITS(parent); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm21 AS SELECT * FROM parent; +ERROR: inheritance parent is not supported on incrementally maintainable materialized view +ROLLBACK; +-- UNION statement is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm22 AS SELECT i,j FROM mv_base_a UNION ALL SELECT i,k FROM mv_base_b;; +ERROR: UNION/INTERSECT/EXCEPT statements are not supported on incrementally maintainable materialized view +-- empty target list is not allowed with IVM +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm25 AS SELECT FROM mv_base_a; +ERROR: empty target list is not supported on incrementally maintainable materialized view +-- FOR UPDATE/SHARE is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm26 AS SELECT i,j FROM mv_base_a FOR UPDATE; +ERROR: FOR UPDATE/SHARE clause is not supported on incrementally maintainable materialized view +-- tartget list cannot contain ivm column that start with '__ivm' +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm28 AS SELECT i AS "__ivm_count__" FROM mv_base_a; +ERROR: column name __ivm_count__ is not supported on incrementally maintainable materialized view +-- expressions specified in GROUP BY must appear in the target list. +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm29 AS SELECT COUNT(i) FROM mv_base_a GROUP BY i; +ERROR: GROUP BY expression not appearing in select list is not supported on incrementally maintainable materialized view +-- experssions containing an aggregate is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm30 AS SELECT sum(i)*0.5 FROM mv_base_a; +ERROR: expression containing an aggregate in it is not supported on incrementally maintainable materialized view +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm31 AS SELECT sum(i)/sum(j) FROM mv_base_a; +ERROR: expression containing an aggregate in it is not supported on incrementally maintainable materialized view +-- VALUES is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_only_values1 AS values(1); +ERROR: VALUES is not supported on incrementally maintainable materialized view +-- views containing base tables with Row Level Security +DROP USER IF EXISTS ivm_admin; +NOTICE: role "ivm_admin" does not exist, skipping +DROP USER IF EXISTS ivm_user; +NOTICE: role "ivm_user" does not exist, skipping +CREATE USER ivm_admin; +CREATE USER ivm_user; +--- create a table with RLS +SET SESSION AUTHORIZATION ivm_admin; +CREATE TABLE rls_tbl(id int, data text, owner name); +INSERT INTO rls_tbl VALUES + (1,'foo','ivm_user'), + (2,'bar','postgres'); +CREATE TABLE num_tbl(id int, num text); +INSERT INTO num_tbl VALUES + (1,'one'), + (2,'two'), + (3,'three'), + (4,'four'), + (5,'five'), + (6,'six'); +--- Users can access only their own rows +CREATE POLICY rls_tbl_policy ON rls_tbl FOR SELECT TO PUBLIC USING(owner = current_user); +ALTER TABLE rls_tbl ENABLE ROW LEVEL SECURITY; +GRANT ALL on rls_tbl TO PUBLIC; +GRANT ALL on num_tbl TO PUBLIC; +--- create a view owned by ivm_user +SET SESSION AUTHORIZATION ivm_user; +CREATE INCREMENTAL MATERIALIZED VIEW ivm_rls AS SELECT * FROM rls_tbl; +NOTICE: could not create an index on materialized view "ivm_rls" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; + id | data | owner +----+------+---------- + 1 | foo | ivm_user +(1 row) + +RESET SESSION AUTHORIZATION; +--- inserts rows owned by different users +INSERT INTO rls_tbl VALUES + (3,'baz','ivm_user'), + (4,'qux','postgres'); +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; + id | data | owner +----+------+---------- + 1 | foo | ivm_user + 3 | baz | ivm_user +(2 rows) + +--- combination of diffent kinds of commands +WITH + i AS (INSERT INTO rls_tbl VALUES(5,'quux','postgres'), (6,'corge','ivm_user')), + u AS (UPDATE rls_tbl SET owner = 'postgres' WHERE id = 1), + u2 AS (UPDATE rls_tbl SET owner = 'ivm_user' WHERE id = 2) +SELECT; +-- +(1 row) + +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; + id | data | owner +----+-------+---------- + 2 | bar | ivm_user + 3 | baz | ivm_user + 6 | corge | ivm_user +(3 rows) + +--- +SET SESSION AUTHORIZATION ivm_user; +CREATE INCREMENTAL MATERIALIZED VIEW ivm_rls2 AS SELECT * FROM rls_tbl JOIN num_tbl USING(id); +NOTICE: could not create an index on materialized view "ivm_rls2" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +RESET SESSION AUTHORIZATION; +WITH + x AS (UPDATE rls_tbl SET data = data || '_2' where id in (3,4)), + y AS (UPDATE num_tbl SET num = num || '_2' where id in (3,4)) +SELECT; +-- +(1 row) + +SELECT * FROM ivm_rls2 ORDER BY 1,2,3; + id | data | owner | num +----+-------+----------+--------- + 2 | bar | ivm_user | two + 3 | baz_2 | ivm_user | three_2 + 6 | corge | ivm_user | six +(3 rows) + +-- automatic index creation +BEGIN; +CREATE TABLE base_a (i int primary key, j int); +CREATE TABLE base_b (i int primary key, j int); +--- group by: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx1 AS SELECT i, sum(j) FROM base_a GROUP BY i; +NOTICE: created index "mv_idx1_index" on materialized view "mv_idx1" +--- distinct: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx2 AS SELECT DISTINCT j FROM base_a; +NOTICE: created index "mv_idx2_index" on materialized view "mv_idx2" +--- with all pkey columns: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx3(i_a, i_b) AS SELECT a.i, b.i FROM base_a a, base_b b; +NOTICE: created index "mv_idx3_index" on materialized view "mv_idx3" +--- missing some pkey columns: no index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx4 AS SELECT j FROM base_a; +NOTICE: could not create an index on materialized view "mv_idx4" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx5 AS SELECT a.i, b.j FROM base_a a, base_b b; +NOTICE: could not create an index on materialized view "mv_idx5" automatically +DETAIL: This target list does not have all the primary key columns, or this view does not contain GROUP BY or DISTINCT clause. +HINT: Create an index on the materialized view for efficient incremental maintenance. +-- cleanup +DROP TABLE rls_tbl CASCADE; +NOTICE: drop cascades to 2 other objects +DETAIL: drop cascades to materialized view ivm_rls +drop cascades to materialized view ivm_rls2 +DROP TABLE num_tbl CASCADE; +DROP USER ivm_user; +DROP USER ivm_admin; +DROP TABLE mv_base_b CASCADE; +NOTICE: drop cascades to 3 other objects +DETAIL: drop cascades to materialized view mv_ivm_1 +drop cascades to view b_view +drop cascades to materialized view b_mview +DROP TABLE mv_base_a CASCADE; diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule index cf46fa3359..013f5db7fb 100644 --- a/src/test/regress/parallel_schedule +++ b/src/test/regress/parallel_schedule @@ -78,7 +78,7 @@ test: brin_bloom brin_multi # psql depends on create_am # amutils depends on geometry, create_index_spgist, hash_index, brin # ---------- -test: create_table_like alter_generic alter_operator misc async dbsize merge misc_functions sysviews tsrf tid tidscan tidrangescan collate.icu.utf8 incremental_sort create_role +test: create_table_like alter_generic alter_operator misc async dbsize merge misc_functions sysviews tsrf tid tidscan tidrangescan collate.icu.utf8 incremental_sort create_role incremental_matview # collate.*.utf8 tests cannot be run in parallel with each other test: rules psql psql_crosstab amutils stats_ext collate.linux.utf8 collate.windows.win1252 diff --git a/src/test/regress/sql/incremental_matview.sql b/src/test/regress/sql/incremental_matview.sql new file mode 100644 index 0000000000..63e26da767 --- /dev/null +++ b/src/test/regress/sql/incremental_matview.sql @@ -0,0 +1,534 @@ +-- create a table to use as a basis for views and materialized views in various combinations +CREATE TABLE mv_base_a (i int, j int); +INSERT INTO mv_base_a VALUES + (1,10), + (2,20), + (3,30), + (4,40), + (5,50); +CREATE TABLE mv_base_b (i int, k int); +INSERT INTO mv_base_b VALUES + (1,101), + (2,102), + (3,103), + (4,104); + +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_1 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) WITH NO DATA; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; +REFRESH MATERIALIZED VIEW mv_ivm_1; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + +-- REFRESH WITH NO DATA +BEGIN; +CREATE FUNCTION dummy_ivm_trigger_func() RETURNS TRIGGER AS $$ + BEGIN + RETURN NULL; + END +$$ language plpgsql; + +CREATE CONSTRAINT TRIGGER dummy_ivm_trigger AFTER INSERT +ON mv_base_a FROM mv_ivm_1 FOR EACH ROW +EXECUTE PROCEDURE dummy_ivm_trigger_func(); + +SELECT COUNT(*) +FROM pg_depend pd INNER JOIN pg_trigger pt ON pd.objid = pt.oid +WHERE pd.classid = 'pg_trigger'::regclass AND pd.refobjid = 'mv_ivm_1'::regclass; + +REFRESH MATERIALIZED VIEW mv_ivm_1 WITH NO DATA; + +SELECT COUNT(*) +FROM pg_depend pd INNER JOIN pg_trigger pt ON pd.objid = pt.oid +WHERE pd.classid = 'pg_trigger'::regclass AND pd.refobjid = 'mv_ivm_1'::regclass; +ROLLBACK; + +-- immediate maintenance +BEGIN; +INSERT INTO mv_base_b VALUES(5,105); +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; +UPDATE mv_base_a SET j = 0 WHERE i = 1; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; +DELETE FROM mv_base_b WHERE (i,k) = (5,105); +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; +ROLLBACK; +SELECT * FROM mv_ivm_1 ORDER BY 1,2,3; + +-- rename of IVM columns +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_rename AS SELECT DISTINCT * FROM mv_base_a; +ALTER MATERIALIZED VIEW mv_ivm_rename RENAME COLUMN __ivm_count__ TO xxx; +DROP MATERIALIZED VIEW mv_ivm_rename; + +-- unique index on IVM columns +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_unique AS SELECT DISTINCT * FROM mv_base_a; +CREATE UNIQUE INDEX ON mv_ivm_unique(__ivm_count__); +CREATE UNIQUE INDEX ON mv_ivm_unique((__ivm_count__)); +CREATE UNIQUE INDEX ON mv_ivm_unique((__ivm_count__ + 1)); +DROP MATERIALIZED VIEW mv_ivm_unique; + +-- TRUNCATE a base table in join views +BEGIN; +TRUNCATE mv_base_a; +SELECT * FROM mv_ivm_1; +ROLLBACK; + +BEGIN; +TRUNCATE mv_base_b; +SELECT * FROM mv_ivm_1; +ROLLBACK; + +-- some query syntax +BEGIN; +CREATE FUNCTION ivm_func() RETURNS int LANGUAGE 'sql' + AS 'SELECT 1' IMMUTABLE; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_func AS SELECT * FROM ivm_func(); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_no_tbl AS SELECT 1; +ROLLBACK; + +-- result of materialized view have DISTINCT clause or the duplicate result. +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_duplicate AS SELECT j FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_distinct AS SELECT DISTINCT j FROM mv_base_a; +INSERT INTO mv_base_a VALUES(6,20); +SELECT * FROM mv_ivm_duplicate ORDER BY 1; +SELECT * FROM mv_ivm_distinct ORDER BY 1; +DELETE FROM mv_base_a WHERE (i,j) = (2,20); +SELECT * FROM mv_ivm_duplicate ORDER BY 1; +SELECT * FROM mv_ivm_distinct ORDER BY 1; +ROLLBACK; + +-- support SUM(), COUNT() and AVG() aggregate functions +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(i), AVG(j) FROM mv_base_a GROUP BY i; +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; +INSERT INTO mv_base_a VALUES(2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; +UPDATE mv_base_a SET j = 200 WHERE (i,j) = (2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; +DELETE FROM mv_base_a WHERE (i,j) = (2,200); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3,4; +ROLLBACK; + +-- support COUNT(*) aggregate function +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3; +INSERT INTO mv_base_a VALUES(2,100); +SELECT * FROM mv_ivm_agg ORDER BY 1,2,3; +ROLLBACK; + +-- TRUNCATE a base table in aggregate views +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg AS SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; +TRUNCATE mv_base_a; +SELECT sum, count FROM mv_ivm_agg; +SELECT i, SUM(j), COUNT(*) FROM mv_base_a GROUP BY i; +ROLLBACK; + +-- support aggregate functions without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_group AS SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; +SELECT * FROM mv_ivm_group ORDER BY 1; +INSERT INTO mv_base_a VALUES(6,60); +SELECT * FROM mv_ivm_group ORDER BY 1; +DELETE FROM mv_base_a; +SELECT * FROM mv_ivm_group ORDER BY 1; +ROLLBACK; + +-- TRUNCATE a base table in aggregate views without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_group AS SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; +TRUNCATE mv_base_a; +SELECT sum, count, avg FROM mv_ivm_group; +SELECT SUM(j), COUNT(j), AVG(j) FROM mv_base_a; +ROLLBACK; + +-- resolved issue: When use AVG() function and values is indivisible, result of AVG() is incorrect. +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_avg_bug AS SELECT i, SUM(j), COUNT(j), AVG(j) FROM mv_base_A GROUP BY i; +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; +INSERT INTO mv_base_a VALUES + (1,0), + (1,0), + (2,30), + (2,30); +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; +DELETE FROM mv_base_a WHERE (i,j) = (1,0); +DELETE FROM mv_base_a WHERE (i,j) = (2,30); +SELECT * FROM mv_ivm_avg_bug ORDER BY 1,2,3; +ROLLBACK; + +-- support MIN(), MAX() aggregate functions +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min_max AS SELECT i, MIN(j), MAX(j) FROM mv_base_a GROUP BY i; +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; +INSERT INTO mv_base_a VALUES + (1,11), (1,12), + (2,21), (2,22), + (3,31), (3,32), + (4,41), (4,42), + (5,51), (5,52); +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; +DELETE FROM mv_base_a WHERE (i,j) IN ((1,10), (2,21), (3,32)); +SELECT * FROM mv_ivm_min_max ORDER BY 1,2,3; +ROLLBACK; + +-- support MIN(), MAX() aggregate functions without GROUP clause +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min_max AS SELECT MIN(j), MAX(j) FROM mv_base_a; +SELECT * FROM mv_ivm_min_max; +INSERT INTO mv_base_a VALUES + (0,0), (6,60), (7,70); +SELECT * FROM mv_ivm_min_max; +DELETE FROM mv_base_a WHERE (i,j) IN ((0,0), (7,70)); +SELECT * FROM mv_ivm_min_max; +DELETE FROM mv_base_a; +SELECT * FROM mv_ivm_min_max; +ROLLBACK; + +-- Test MIN/MAX after search_path change +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_min AS SELECT MIN(j) FROM mv_base_a; +SELECT * FROM mv_ivm_min; + +CREATE SCHEMA myschema; +GRANT ALL ON SCHEMA myschema TO public; +CREATE TABLE myschema.mv_base_a (j int); +INSERT INTO myschema.mv_base_a VALUES (1); + +DELETE FROM mv_base_a WHERE (i,j) = (1,10); +SELECT * FROM mv_ivm_min; + +SET search_path TO myschema,public,pg_catalog; +DELETE FROM public.mv_base_a WHERE (i,j) = (2,20); +SELECT * FROM mv_ivm_min; +ROLLBACK; + +-- aggregate views with column names specified +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +INSERT INTO mv_base_a VALUES (1,100), (2,200), (3,300); +UPDATE mv_base_a SET j = 2000 WHERE (i,j) = (2,20); +DELETE FROM mv_base_a WHERE (i,j) = (3,30); +SELECT * FROM mv_ivm_agg ORDER BY 1,2; +ROLLBACK; +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a,b) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +INSERT INTO mv_base_a VALUES (1,100), (2,200), (3,300); +UPDATE mv_base_a SET j = 2000 WHERE (i,j) = (2,20); +DELETE FROM mv_base_a WHERE (i,j) = (3,30); +SELECT * FROM mv_ivm_agg ORDER BY 1,2; +ROLLBACK; +BEGIN; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_agg(a,b,c) AS SELECT i, SUM(j) FROM mv_base_a GROUP BY i; +ROLLBACK; + +-- support self join view and multiple change on the same table +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (1, 10), (2, 20), (3, 30); +CREATE INCREMENTAL MATERIALIZED VIEW mv_self(v1, v2) AS + SELECT t1.v, t2.v FROM base_t AS t1 JOIN base_t AS t2 ON t1.i = t2.i; +SELECT * FROM mv_self ORDER BY v1; +INSERT INTO base_t VALUES (4,40); +DELETE FROM base_t WHERE i = 1; +UPDATE base_t SET v = v*10 WHERE i=2; +SELECT * FROM mv_self ORDER BY v1; +WITH + ins_t1 AS (INSERT INTO base_t VALUES (5,50) RETURNING 1), + ins_t2 AS (INSERT INTO base_t VALUES (6,60) RETURNING 1), + upd_t AS (UPDATE base_t SET v = v + 100 RETURNING 1), + dlt_t AS (DELETE FROM base_t WHERE i IN (4,5) RETURNING 1) +SELECT NULL; +SELECT * FROM mv_self ORDER BY v1; + +--- with sub-transactions +SAVEPOINT p1; +INSERT INTO base_t VALUES (7,70); +RELEASE SAVEPOINT p1; +INSERT INTO base_t VALUES (7,77); +SELECT * FROM mv_self ORDER BY v1, v2; + +ROLLBACK; + +-- support simultaneous table changes +BEGIN; +CREATE TABLE base_r (i int, v int); +CREATE TABLE base_s (i int, v int); +INSERT INTO base_r VALUES (1, 10), (2, 20), (3, 30); +INSERT INTO base_s VALUES (1, 100), (2, 200), (3, 300); +CREATE INCREMENTAL MATERIALIZED VIEW mv(v1, v2) AS + SELECT r.v, s.v FROM base_r AS r JOIN base_s AS s USING(i); +SELECT * FROM mv ORDER BY v1; +WITH + ins_r AS (INSERT INTO base_r VALUES (1,11) RETURNING 1), + ins_r2 AS (INSERT INTO base_r VALUES (3,33) RETURNING 1), + ins_s AS (INSERT INTO base_s VALUES (2,222) RETURNING 1), + upd_r AS (UPDATE base_r SET v = v + 1000 WHERE i = 2 RETURNING 1), + dlt_s AS (DELETE FROM base_s WHERE i = 3 RETURNING 1) +SELECT NULL; +SELECT * FROM mv ORDER BY v1; +ROLLBACK; + +-- support foreign reference constraints +BEGIN; +CREATE TABLE ri1 (i int PRIMARY KEY); +CREATE TABLE ri2 (i int PRIMARY KEY REFERENCES ri1(i) ON UPDATE CASCADE ON DELETE CASCADE, v int); +INSERT INTO ri1 VALUES (1),(2),(3); +INSERT INTO ri2 VALUES (1),(2),(3); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ri(i1, i2) AS + SELECT ri1.i, ri2.i FROM ri1 JOIN ri2 USING(i); +SELECT * FROM mv_ri ORDER BY i1; +UPDATE ri1 SET i=10 where i=1; +DELETE FROM ri1 WHERE i=2; +SELECT * FROM mv_ri ORDER BY i2; +ROLLBACK; + +-- views including NULL +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (1,10),(2, NULL); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT * FROM base_t; +SELECT * FROM mv ORDER BY i; +UPDATE base_t SET v = 20 WHERE i = 2; +SELECT * FROM mv ORDER BY i; +ROLLBACK; + +BEGIN; +CREATE TABLE base_t (i int); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT * FROM base_t; +SELECT * FROM mv ORDER BY i; +INSERT INTO base_t VALUES (1),(NULL); +SELECT * FROM mv ORDER BY i; +ROLLBACK; + +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (NULL, 1), (NULL, 2), (1, 10), (1, 20); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT i, sum(v) FROM base_t GROUP BY i; +SELECT * FROM mv ORDER BY i; +UPDATE base_t SET v = v * 10; +SELECT * FROM mv ORDER BY i; +ROLLBACK; + +BEGIN; +CREATE TABLE base_t (i int, v int); +INSERT INTO base_t VALUES (NULL, 1), (NULL, 2), (NULL, 3), (NULL, 4), (NULL, 5); +CREATE INCREMENTAL MATERIALIZED VIEW mv AS SELECT i, min(v), max(v) FROM base_t GROUP BY i; +SELECT * FROM mv ORDER BY i; +DELETE FROM base_t WHERE v = 1; +SELECT * FROM mv ORDER BY i; +DELETE FROM base_t WHERE v = 3; +SELECT * FROM mv ORDER BY i; +DELETE FROM base_t WHERE v = 5; +SELECT * FROM mv ORDER BY i; +ROLLBACK; + +-- IMMV containing user defined type +BEGIN; + +CREATE TYPE mytype; +CREATE FUNCTION mytype_in(cstring) + RETURNS mytype AS 'int4in' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE FUNCTION mytype_out(mytype) + RETURNS cstring AS 'int4out' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE TYPE mytype ( + LIKE = int4, + INPUT = mytype_in, + OUTPUT = mytype_out +); + +CREATE FUNCTION mytype_eq(mytype, mytype) + RETURNS bool AS 'int4eq' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE FUNCTION mytype_lt(mytype, mytype) + RETURNS bool AS 'int4lt' + LANGUAGE INTERNAL STRICT IMMUTABLE; +CREATE FUNCTION mytype_cmp(mytype, mytype) + RETURNS integer AS 'btint4cmp' + LANGUAGE INTERNAL STRICT IMMUTABLE; + +CREATE OPERATOR = ( + leftarg = mytype, rightarg = mytype, + procedure = mytype_eq); +CREATE OPERATOR < ( + leftarg = mytype, rightarg = mytype, + procedure = mytype_lt); + +CREATE OPERATOR CLASS mytype_ops + DEFAULT FOR TYPE mytype USING btree AS + OPERATOR 1 <, + OPERATOR 3 = , + FUNCTION 1 mytype_cmp(mytype,mytype); + +CREATE TABLE t_mytype (x mytype); +CREATE INCREMENTAL MATERIALIZED VIEW mv_mytype AS + SELECT * FROM t_mytype; +INSERT INTO t_mytype VALUES ('1'::mytype); +SELECT * FROM mv_mytype; + +ROLLBACK; + +-- outer join is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv(a,b) AS SELECT a.i, b.i FROM mv_base_a a LEFT JOIN mv_base_b b ON a.i=b.i; +-- CTE is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv AS + WITH b AS ( SELECT * FROM mv_base_b) SELECT a.i,a.j FROM mv_base_a a, b WHERE a.i = b.i; +-- contain system column +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm01 AS SELECT i,j,xmin FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm02 AS SELECT i,j FROM mv_base_a WHERE xmin = '610'; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm04 AS SELECT i,j,xmin::text AS x_min FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm06 AS SELECT i,j,xidsend(xmin) AS x_min FROM mv_base_a; +-- contain subquery +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm03 AS SELECT i,j FROM mv_base_a WHERE i IN (SELECT i FROM mv_base_b WHERE k < 103 ); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm04 AS SELECT a.i,a.j FROM mv_base_a a, (SELECT * FROM mv_base_b) b WHERE a.i = b.i; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm05 AS SELECT i,j, (SELECT k FROM mv_base_b b WHERE a.i = b.i) FROM mv_base_a a; +-- contain ORDER BY +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm07 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) ORDER BY i,j,k; +-- contain HAVING +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm08 AS SELECT i,j,k FROM mv_base_a a INNER JOIN mv_base_b b USING(i) GROUP BY i,j,k HAVING SUM(i) > 5; + +-- contain view or materialized view +CREATE VIEW b_view AS SELECT i,k FROM mv_base_b; +CREATE MATERIALIZED VIEW b_mview AS SELECT i,k FROM mv_base_b; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm07 AS SELECT a.i,a.j FROM mv_base_a a,b_view b WHERE a.i = b.i; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm08 AS SELECT a.i,a.j FROM mv_base_a a,b_mview b WHERE a.i = b.i; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm09 AS SELECT a.i,a.j FROM mv_base_a a, (SELECT i, COUNT(*) FROM mv_base_b GROUP BY i) b WHERE a.i = b.i; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm10 AS SELECT a.i,a.j FROM mv_base_a a WHERE EXISTS(SELECT 1 FROM mv_base_b b WHERE a.i = b.i) OR a.i > 5; + +-- contain mutable functions +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm12 AS SELECT i,j FROM mv_base_a WHERE i = random()::int; + +-- LIMIT/OFFSET is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm13 AS SELECT i,j FROM mv_base_a LIMIT 10 OFFSET 5; + +-- DISTINCT ON is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm14 AS SELECT DISTINCT ON(i) i, j FROM mv_base_a; + +-- TABLESAMPLE clause is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm15 AS SELECT i, j FROM mv_base_a TABLESAMPLE SYSTEM(50); + +-- window functions are not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm16 AS SELECT *, cume_dist() OVER (ORDER BY i) AS rank FROM mv_base_a; + +-- aggregate function with some options is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm17 AS SELECT COUNT(*) FILTER(WHERE i < 3) FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm18 AS SELECT COUNT(DISTINCT i) FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm19 AS SELECT array_agg(j ORDER BY i DESC) FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm20 AS SELECT i,SUM(j) FROM mv_base_a GROUP BY GROUPING SETS((i),()); + +-- inheritance parent is not supported +BEGIN; +CREATE TABLE parent (i int, v int); +CREATE TABLE child_a(options text) INHERITS(parent); +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm21 AS SELECT * FROM parent; +ROLLBACK; + +-- UNION statement is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm22 AS SELECT i,j FROM mv_base_a UNION ALL SELECT i,k FROM mv_base_b;; + +-- empty target list is not allowed with IVM +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm25 AS SELECT FROM mv_base_a; + +-- FOR UPDATE/SHARE is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm26 AS SELECT i,j FROM mv_base_a FOR UPDATE; + +-- tartget list cannot contain ivm column that start with '__ivm' +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm28 AS SELECT i AS "__ivm_count__" FROM mv_base_a; + +-- expressions specified in GROUP BY must appear in the target list. +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm29 AS SELECT COUNT(i) FROM mv_base_a GROUP BY i; + +-- experssions containing an aggregate is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm30 AS SELECT sum(i)*0.5 FROM mv_base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm31 AS SELECT sum(i)/sum(j) FROM mv_base_a; + +-- VALUES is not supported +CREATE INCREMENTAL MATERIALIZED VIEW mv_ivm_only_values1 AS values(1); + +-- views containing base tables with Row Level Security +DROP USER IF EXISTS ivm_admin; +DROP USER IF EXISTS ivm_user; +CREATE USER ivm_admin; +CREATE USER ivm_user; + +--- create a table with RLS +SET SESSION AUTHORIZATION ivm_admin; +CREATE TABLE rls_tbl(id int, data text, owner name); +INSERT INTO rls_tbl VALUES + (1,'foo','ivm_user'), + (2,'bar','postgres'); +CREATE TABLE num_tbl(id int, num text); +INSERT INTO num_tbl VALUES + (1,'one'), + (2,'two'), + (3,'three'), + (4,'four'), + (5,'five'), + (6,'six'); + +--- Users can access only their own rows +CREATE POLICY rls_tbl_policy ON rls_tbl FOR SELECT TO PUBLIC USING(owner = current_user); +ALTER TABLE rls_tbl ENABLE ROW LEVEL SECURITY; +GRANT ALL on rls_tbl TO PUBLIC; +GRANT ALL on num_tbl TO PUBLIC; + +--- create a view owned by ivm_user +SET SESSION AUTHORIZATION ivm_user; + +CREATE INCREMENTAL MATERIALIZED VIEW ivm_rls AS SELECT * FROM rls_tbl; +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; +RESET SESSION AUTHORIZATION; + +--- inserts rows owned by different users +INSERT INTO rls_tbl VALUES + (3,'baz','ivm_user'), + (4,'qux','postgres'); +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; + +--- combination of diffent kinds of commands +WITH + i AS (INSERT INTO rls_tbl VALUES(5,'quux','postgres'), (6,'corge','ivm_user')), + u AS (UPDATE rls_tbl SET owner = 'postgres' WHERE id = 1), + u2 AS (UPDATE rls_tbl SET owner = 'ivm_user' WHERE id = 2) +SELECT; +SELECT id, data, owner FROM ivm_rls ORDER BY 1,2,3; + +--- +SET SESSION AUTHORIZATION ivm_user; +CREATE INCREMENTAL MATERIALIZED VIEW ivm_rls2 AS SELECT * FROM rls_tbl JOIN num_tbl USING(id); +RESET SESSION AUTHORIZATION; + +WITH + x AS (UPDATE rls_tbl SET data = data || '_2' where id in (3,4)), + y AS (UPDATE num_tbl SET num = num || '_2' where id in (3,4)) +SELECT; +SELECT * FROM ivm_rls2 ORDER BY 1,2,3; + +-- automatic index creation +BEGIN; +CREATE TABLE base_a (i int primary key, j int); +CREATE TABLE base_b (i int primary key, j int); + +--- group by: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx1 AS SELECT i, sum(j) FROM base_a GROUP BY i; + +--- distinct: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx2 AS SELECT DISTINCT j FROM base_a; + +--- with all pkey columns: create an index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx3(i_a, i_b) AS SELECT a.i, b.i FROM base_a a, base_b b; + +--- missing some pkey columns: no index +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx4 AS SELECT j FROM base_a; +CREATE INCREMENTAL MATERIALIZED VIEW mv_idx5 AS SELECT a.i, b.j FROM base_a a, base_b b; + +-- cleanup + +DROP TABLE rls_tbl CASCADE; +DROP TABLE num_tbl CASCADE; +DROP USER ivm_user; +DROP USER ivm_admin; + +DROP TABLE mv_base_b CASCADE; +DROP TABLE mv_base_a CASCADE; -- 2.25.1 --Multipart=_Thu__1_Jun_2023_23_59_09_+0900_/G5+8nG46.f1T42K Content-Type: text/x-diff; name="v28-0011-Add-documentations-about-Incremental-View-Mainte.patch" Content-Disposition: attachment; filename="v28-0011-Add-documentations-about-Incremental-View-Mainte.patch" Content-Transfer-Encoding: 7bit ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v1] solaris: Use versioning scripts instead of -Bsymbolic @ 2022-08-23 08:50 Andres Freund <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Andres Freund @ 2022-08-23 08:50 UTC (permalink / raw) -Bsymbolic causes a lot of "ld: warning: symbol referencing errors" warnings. It's quite easy to add the symbol versioning script, we just need a slightly different parameter name. Discussion: https://postgr.es/m/[email protected] --- src/Makefile.shlib | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/src/Makefile.shlib b/src/Makefile.shlib index 2af6192f0f3..3202b2e67de 100644 --- a/src/Makefile.shlib +++ b/src/Makefile.shlib @@ -188,10 +188,15 @@ ifeq ($(PORTNAME), linux) endif ifeq ($(PORTNAME), solaris) - LINK.shared = $(COMPILER) -shared -Wl,-Bsymbolic + LINK.shared = $(COMPILER) -shared ifdef soname LINK.shared += -Wl,-soname,$(soname) endif + BUILD.exports = ( echo '{ global:'; $(AWK) '/^[^\#]/ {printf "%s;\n",$$1}' $<; echo ' local: *; };' ) >$@ + exports_file = $(SHLIB_EXPORTS:%.txt=%.list) + ifneq (,$(exports_file)) + LINK.shared += -Wl,-M$(exports_file) + endif endif ifeq ($(PORTNAME), cygwin) -- 2.37.0.3.g30cc8d0f14 --aym2ahtpoerge4dw-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 274ad900c65..5188244dfd6 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -993,6 +993,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0f6b5df322c..23511a0905a 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --3n3vqr5y2jhknrqj Content-Type: text/x-diff; charset=utf-8 Content-Disposition: attachment; filename="v49-0007-WIP-add-max_repack_replication_slots.patch" ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 7291b4180ab..0103bc10bd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --cldir6umqisr673d-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 4fd10a5e7b3..73407c383da 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1086,6 +1086,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --=-=-=-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 73 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 6 +- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 138 insertions(+), 22 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..79c052d3607 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXid %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 065316ee866..77c206ff944 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1084,6 +1084,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 0f913897acc..f1a03e6a754 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,13 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2687,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2702,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2711,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2732,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2754,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2773,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2826,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2848,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..5c4121fcbc8 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1372,7 +1375,8 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) /* array of TransactionIds */ if (xlrec.xcnt > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index bf3094f0f7d..120370f459d 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -67,10 +67,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index c63a4f2cc6a..0bd28e270f1 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -124,10 +124,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --nzintiedl6o4kcyp-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/cluster.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/access/xlog_internal.h | 2 +- src/include/storage/proc.h | 3 +- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 10 files changed, 140 insertions(+), 23 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 018e9f7b013..423cea26b0b 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1075,6 +1075,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..b02ec4e8c3e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index f3ad90c7c7a..c785a3db59f 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1188,6 +1188,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1358,10 +1359,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1370,9 +1373,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 629ac3a7d3e..1730b07810f 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -31,7 +31,7 @@ /* * Each page of XLOG file has a header like this: */ -#define XLOG_PAGE_MAGIC 0xD11D /* can be used as WAL version indicator */ +#define XLOG_PAGE_MAGIC 0xD11E /* can be used as WAL version indicator */ typedef struct XLogPageHeaderData { diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 1dad125706e..c6693b93a65 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -69,10 +69,11 @@ struct XidCache #define PROC_AFFECTS_ALL_HORIZONS 0x20 /* this proc's xmin must be * included in vacuum horizons * in all databases */ +#define PROC_IN_REPACK 0x40 /* currently REPACK (CONCURRENTLY) */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ - (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND) + (PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | PROC_IN_REPACK) /* * Xmin-related flags. Make sure any flags that affect how the process' Xmin diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --gwom7bl7ogtszo4k-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 3acb2e1d2ba..74faa4f5582 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --2fxmamo6mu2qbxgv-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 51 +++++++++----- src/backend/storage/ipc/procarray.c | 75 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 2 + src/include/storage/standbydefs.h | 2 + 8 files changed, 144 insertions(+), 27 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..58ebecf4927 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 2c1c6f88b74..f1cbdee7618 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8483,6 +8484,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 781a471fc1d..29ba49744eb 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is o.k. + * for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..f2008c6877c 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid - * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This - * means all transactions starting afterwards have enough information to - * be decoded. Switch to FULL_SNAPSHOT. + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * BUILDING_SNAPSHOT. This means all transactions starting afterwards + * have enough information to be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,15 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is - * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all - * transactions that are currently in progress have a catalog snapshot, - * and all their changes have been collected. Switch to CONSISTENT. + * In FULL_SNAPSHOT state, and this xl_running_xacts' + * oldestRunningXidLogical is >= than nextXid from when we switched to + * FULL_SNAPSHOT. This means all transactions that are currently in + * progress have a catalog snapshot, and all their changes have been + * collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1460,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..033ae68b57e 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,25 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, + count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2673,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2689,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2704,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2713,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2734,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2756,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2775,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2828,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2850,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..fe4dcdc3def 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,12 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --qfkt2ktdpcfeypib-- ^ permalink raw reply [nested|flat] 426+ messages in thread
* [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). @ 2026-03-17 19:22 Antonin Houska <[email protected]> 0 siblings, 0 replies; 426+ messages in thread From: Antonin Houska @ 2026-03-17 19:22 UTC (permalink / raw) During logical decoding, we need to know if particular transaction performed catalog changes because catalog information is needed to construct heap tuples. To be sure that we have enough information of each transaction, the logical decoding cannot start before all the already running transactions have completed. The problem with REPACK (CONCURRENTLY) is that it has XID assigned and writes WAL records marked with it. Thus if another backend runs REPACK (CONCURRENTLY) and tries to setup the logical decoding, it has to wait for the completion of all the other transactions involved in REPACK (CONCURRENTLY). However, REPACK (CONCURRENTLY) does not perform any catalog changes relevant to logical decoding, so the other backends executing this command can ignore it. This patch implements it by adding information about transactions executing the command to the xl_running_xacts WAL record, and by teaching the snapshot builder to use the information. --- src/backend/access/rmgrdesc/standbydesc.c | 15 ++++- src/backend/access/transam/xlog.c | 2 + src/backend/commands/repack.c | 16 +++++ src/backend/replication/logical/snapbuild.c | 38 ++++++++--- src/backend/storage/ipc/procarray.c | 74 +++++++++++++++++++-- src/backend/storage/ipc/standby.c | 8 ++- src/include/storage/standby.h | 3 + src/include/storage/standbydefs.h | 2 + 8 files changed, 137 insertions(+), 21 deletions(-) diff --git a/src/backend/access/rmgrdesc/standbydesc.c b/src/backend/access/rmgrdesc/standbydesc.c index 0a291354ae2..b96b73d51d3 100644 --- a/src/backend/access/rmgrdesc/standbydesc.c +++ b/src/backend/access/rmgrdesc/standbydesc.c @@ -21,10 +21,11 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) { int i; - appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u", + appendStringInfo(buf, "nextXid %u latestCompletedXid %u oldestRunningXid %u oldestRunningXidLogical %u", xlrec->nextXid, xlrec->latestCompletedXid, - xlrec->oldestRunningXid); + xlrec->oldestRunningXid, + xlrec->oldestRunningXidLogical); if (xlrec->xcnt > 0) { appendStringInfo(buf, "; %d xacts:", xlrec->xcnt); @@ -41,6 +42,16 @@ standby_desc_running_xacts(StringInfo buf, xl_running_xacts *xlrec) for (i = 0; i < xlrec->subxcnt; i++) appendStringInfo(buf, " %u", xlrec->xids[xlrec->xcnt + i]); } + + if (xlrec->xcnt_repack > 0) + { + TransactionId *xids_repack; + + appendStringInfo(buf, "; %d xacts_repack:", xlrec->xcnt_repack); + xids_repack = xlrec->xids + xlrec->xcnt + xlrec->subxcnt; + for (i = 0; i < xlrec->xcnt_repack; i++) + appendStringInfo(buf, " %u", xids_repack[i]); + } } void diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f5c9a34374d..d050b0b1444 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -5912,6 +5912,7 @@ StartupXLOG(void) * subxids are listed with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); @@ -8477,6 +8478,7 @@ xlog_redo(XLogReaderState *record) * with their parent prepared transactions. */ running.xcnt = nxids; + running.xcnt_repack = 0; running.subxcnt = 0; running.subxid_status = SUBXIDS_IN_SUBTRANS; running.nextXid = XidFromFullTransactionId(checkPoint.nextXid); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 21ab81c6f7f..42165c5552e 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -988,6 +988,22 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose, if (concurrent) { + /* + * Do not let other backends wait for our completion during their + * setup of logical replication. Unlike logical replication publisher, + * we will have XID assigned, so the other backends - whether + * walsenders involved in logical replication or regular backends + * executing also REPACK (CONCURRENTLY) - would have to wait for our + * completion before they can build their initial snapshot. It is + * o.k. for any decoding backend to ignore us because we do not change + * tuple descriptor of any table, and the data changes we write should + * not be decoded by other backends. + */ + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK; + ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; + LWLockRelease(ProcArrayLock); + /* * The worker needs to be member of the locking group we're the leader * of. We ought to become the leader before the worker starts. The diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 9cf499ce7c6..fbdd4600a2b 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -1172,7 +1172,7 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact * xmin, which looks odd but is correct and actually more efficient, since * we hit fast paths in heapam_visibility.c. */ - builder->xmin = running->oldestRunningXid; + builder->xmin = running->oldestRunningXidLogical; /* Remove transactions we don't need to keep track off anymore */ SnapBuildPurgeOlderTxn(builder); @@ -1188,9 +1188,9 @@ SnapBuildProcessRunningXacts(SnapBuild *builder, XLogRecPtr lsn, xl_running_xact */ xmin = ReorderBufferGetOldestXmin(builder->reorder); if (xmin == InvalidTransactionId) - xmin = running->oldestRunningXid; + xmin = running->oldestRunningXidLogical; elog(DEBUG3, "xmin: %u, xmax: %u, oldest running: %u, oldest xmin: %u", - builder->xmin, builder->xmax, running->oldestRunningXid, xmin); + builder->xmin, builder->xmax, running->oldestRunningXidLogical, xmin); LogicalIncreaseXminForSlot(lsn, xmin); /* @@ -1275,14 +1275,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * have all necessary catalog rows anymore. */ if (TransactionIdIsNormal(builder->initial_xmin_horizon) && - NormalTransactionIdPrecedes(running->oldestRunningXid, + NormalTransactionIdPrecedes(running->oldestRunningXidLogical, builder->initial_xmin_horizon)) { ereport(DEBUG1, errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", LSN_FORMAT_ARGS(lsn)), errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid)); + builder->initial_xmin_horizon, running->oldestRunningXidLogical)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1299,7 +1299,7 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn * NB: We might have already started to incrementally assemble a snapshot, * so we need to be careful to deal with that. */ - if (running->oldestRunningXid == running->nextXid) + if (running->oldestRunningXidLogical == running->nextXid) { if (!XLogRecPtrIsValid(builder->start_decoding_at) || builder->start_decoding_at <= lsn) @@ -1378,14 +1378,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from BUILDING_SNAPSHOT to FULL_SNAPSHOT. * - * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid + * In BUILDING_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical * is >= than nextXid from when we switched to BUILDING_SNAPSHOT. This * means all transactions starting afterwards have enough information to * be decoded. Switch to FULL_SNAPSHOT. */ else if (builder->state == SNAPBUILD_BUILDING_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_FULL_SNAPSHOT; builder->next_phase_at = running->nextXid; @@ -1402,14 +1402,14 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn /* * c) transition from FULL_SNAPSHOT to CONSISTENT. * - * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXid is + * In FULL_SNAPSHOT state, and this xl_running_xacts' oldestRunningXidLogical is * >= than nextXid from when we switched to FULL_SNAPSHOT. This means all * transactions that are currently in progress have a catalog snapshot, * and all their changes have been collected. Switch to CONSISTENT. */ else if (builder->state == SNAPBUILD_FULL_SNAPSHOT && TransactionIdPrecedesOrEquals(builder->next_phase_at, - running->oldestRunningXid)) + running->oldestRunningXidLogical)) { builder->state = SNAPBUILD_CONSISTENT; builder->next_phase_at = InvalidTransactionId; @@ -1459,6 +1459,24 @@ SnapBuildWaitSnapshot(xl_running_xacts *running, TransactionId cutoff) if (TransactionIdFollows(xid, cutoff)) continue; + /* Do not wait for transactions running REPACK (CONCURRENTLY). */ + if (running->xcnt_repack > 0) + { + TransactionId *xids_repack; + int i; + + xids_repack = running->xids + running->xcnt + running->subxcnt; + + for (i = 0; i < running->xcnt_repack; i++) + { + if (xid == xids_repack[i]) + break; + } + /* Found? */ + if (i < running->xcnt_repack) + continue; + } + XactLockTableWait(xid, NULL, NULL, XLTW_None); } diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index cc207cb56e3..6dd4fa7dcc0 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -2643,15 +2643,24 @@ GetRunningTransactionData(void) RunningTransactions CurrentRunningXacts = &CurrentRunningXactsData; TransactionId latestCompletedXid; TransactionId oldestRunningXid; + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; TransactionId *xids; int index; - int count; + int count, count_repack; int subcount; bool suboverflowed; + TransactionId *xids_repack = NULL; + bool logical_decoding_enabled = IsLogicalDecodingEnabled(); Assert(!RecoveryInProgress()); + /* + * TODO Consider a GUC to reserve certain amount of replication slots for + * REPACK (CONCURRENTLY) and using it here. + */ +#define MAX_REPACK_XIDS 16 + /* * Allocating space for maxProcs xids is usually overkill; numProcs would * be sufficient. But it seems better to do the malloc while not holding @@ -2663,11 +2672,14 @@ GetRunningTransactionData(void) */ if (CurrentRunningXacts->xids == NULL) { + /* FIXME probably fails if logical decoding is enable on-the-fly */ + int nrepack = logical_decoding_enabled ? MAX_REPACK_XIDS : 0; + /* * First call */ CurrentRunningXacts->xids = (TransactionId *) - malloc(TOTAL_MAX_CACHED_SUBXIDS * sizeof(TransactionId)); + malloc((TOTAL_MAX_CACHED_SUBXIDS + nrepack) * sizeof(TransactionId)); if (CurrentRunningXacts->xids == NULL) ereport(ERROR, (errcode(ERRCODE_OUT_OF_MEMORY), @@ -2676,7 +2688,10 @@ GetRunningTransactionData(void) xids = CurrentRunningXacts->xids; - count = subcount = 0; + if (logical_decoding_enabled) + xids_repack = palloc_array(TransactionId, MAX_REPACK_XIDS); + + count = subcount = count_repack = 0; suboverflowed = false; /* @@ -2688,7 +2703,7 @@ GetRunningTransactionData(void) latestCompletedXid = XidFromFullTransactionId(TransamVariables->latestCompletedXid); - oldestDatabaseRunningXid = oldestRunningXid = + oldestDatabaseRunningXid = oldestRunningXid = oldestRunningXidLogical = XidFromFullTransactionId(TransamVariables->nextXid); /* @@ -2697,6 +2712,8 @@ GetRunningTransactionData(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno; + PGPROC *proc; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2716,6 +2733,21 @@ GetRunningTransactionData(void) if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; + if (logical_decoding_enabled && + TransactionIdPrecedes(xid, oldestRunningXidLogical)) + { + /* + * Backends running REPACK concurrently need to be excluded from + * oldestRunningXidLogical, otherwise the snapshot builder cannot + * proceed in building the initial snapshot. + */ + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if ((proc->statusFlags & PROC_IN_CONCURRENT_REPACK) == 0) + oldestRunningXidLogical = xid; + } + /* * Also, update the oldest running xid within the current database. As * fetching pgprocno and PGPROC could cause cache misses, we do cheap @@ -2723,8 +2755,8 @@ GetRunningTransactionData(void) */ if (TransactionIdPrecedes(xid, oldestDatabaseRunningXid)) { - int pgprocno = arrayP->pgprocnos[index]; - PGPROC *proc = &allProcs[pgprocno]; + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; if (proc->databaseId == MyDatabaseId) oldestDatabaseRunningXid = xid; @@ -2742,6 +2774,19 @@ GetRunningTransactionData(void) */ xids[count++] = xid; + + /* + * Collect XIDSs of transactions running REPACK (CONCURRENTLY). + */ + if (logical_decoding_enabled && + count_repack < MAX_REPACK_XIDS) + { + pgprocno = arrayP->pgprocnos[index]; + proc = &allProcs[pgprocno]; + + if (proc->statusFlags & PROC_IN_CONCURRENT_REPACK) + xids_repack[count_repack++] = xid; + } } /* @@ -2782,6 +2827,19 @@ GetRunningTransactionData(void) } } + /* + * Append the XIDs running REPACK (CONCURRENTLY), if any. + * + * XXX Should we sort the array and use bsearch() when using it? + */ + if (count_repack > 0) + { + for (int i = 0; i < count_repack; i++) + xids[count++] = xids_repack[i]; + } + if (xids_repack) + pfree(xids_repack); + /* * It's important *not* to include the limits set by slots here because * snapbuild.c uses oldestRunningXid to manage its xmin horizon. If those @@ -2791,11 +2849,13 @@ GetRunningTransactionData(void) * increases if slots do. */ - CurrentRunningXacts->xcnt = count - subcount; + CurrentRunningXacts->xcnt = count - subcount - count_repack; CurrentRunningXacts->subxcnt = subcount; + CurrentRunningXacts->xcnt_repack = count_repack; CurrentRunningXacts->subxid_status = suboverflowed ? SUBXIDS_IN_SUBTRANS : SUBXIDS_IN_ARRAY; CurrentRunningXacts->nextXid = XidFromFullTransactionId(TransamVariables->nextXid); CurrentRunningXacts->oldestRunningXid = oldestRunningXid; + CurrentRunningXacts->oldestRunningXidLogical = oldestRunningXidLogical; CurrentRunningXacts->oldestDatabaseRunningXid = oldestDatabaseRunningXid; CurrentRunningXacts->latestCompletedXid = latestCompletedXid; diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index de9092fdf5b..fd3e4fe31f3 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1189,6 +1189,7 @@ standby_redo(XLogReaderState *record) RunningTransactionsData running; running.xcnt = xlrec->xcnt; + running.xcnt_repack = xlrec->xcnt_repack; running.subxcnt = xlrec->subxcnt; running.subxid_status = xlrec->subxid_overflow ? SUBXIDS_MISSING : SUBXIDS_IN_ARRAY; running.nextXid = xlrec->nextXid; @@ -1359,10 +1360,12 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRecPtr recptr; xlrec.xcnt = CurrRunningXacts->xcnt; + xlrec.xcnt_repack = CurrRunningXacts->xcnt_repack; xlrec.subxcnt = CurrRunningXacts->subxcnt; xlrec.subxid_overflow = (CurrRunningXacts->subxid_status != SUBXIDS_IN_ARRAY); xlrec.nextXid = CurrRunningXacts->nextXid; xlrec.oldestRunningXid = CurrRunningXacts->oldestRunningXid; + xlrec.oldestRunningXidLogical = CurrRunningXacts->oldestRunningXidLogical; xlrec.latestCompletedXid = CurrRunningXacts->latestCompletedXid; /* Header */ @@ -1371,9 +1374,10 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) XLogRegisterData(&xlrec, MinSizeOfXactRunningXacts); /* array of TransactionIds */ - if (xlrec.xcnt > 0) + if (xlrec.xcnt + xlrec.xcnt_repack > 0) XLogRegisterData(CurrRunningXacts->xids, - (xlrec.xcnt + xlrec.subxcnt) * sizeof(TransactionId)); + (xlrec.xcnt + xlrec.xcnt_repack + xlrec.subxcnt) * + sizeof(TransactionId)); recptr = XLogInsert(RM_STANDBY_ID, XLOG_RUNNING_XACTS); diff --git a/src/include/storage/standby.h b/src/include/storage/standby.h index 6a314c693cd..9e6e7b18786 100644 --- a/src/include/storage/standby.h +++ b/src/include/storage/standby.h @@ -127,10 +127,13 @@ typedef enum typedef struct RunningTransactionsData { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK + * (CONCURRENTLY). */ int subxcnt; /* # of subxact ids in xids[] */ subxids_array_status subxid_status; TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId oldestDatabaseRunningXid; /* same as above, but within the * current database */ TransactionId latestCompletedXid; /* so we can set xmax */ diff --git a/src/include/storage/standbydefs.h b/src/include/storage/standbydefs.h index 231d251fd51..edad609fa9a 100644 --- a/src/include/storage/standbydefs.h +++ b/src/include/storage/standbydefs.h @@ -47,10 +47,12 @@ typedef struct xl_standby_locks typedef struct xl_running_xacts { int xcnt; /* # of xact ids in xids[] */ + int xcnt_repack; /* # of xacts running REPACK (CONCURRENTLY) */ int subxcnt; /* # of subxact ids in xids[] */ bool subxid_overflow; /* snapshot overflowed, subxids missing */ TransactionId nextXid; /* xid from TransamVariables->nextXid */ TransactionId oldestRunningXid; /* *not* oldestXmin */ + TransactionId oldestRunningXidLogical; TransactionId latestCompletedXid; /* so we can set xmax */ TransactionId xids[FLEXIBLE_ARRAY_MEMBER]; -- 2.47.3 --mdsv5bdgfjjj6d77-- ^ permalink raw reply [nested|flat] 426+ messages in thread
end of thread, other threads:[~2026-03-17 19:22 UTC | newest] Thread overview: 426+ messages (download: mbox mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2021-03-10 02:11 [PATCH v28 10/11] Add regression tests for Incremental View Maintenance Takuma Hoshiai <[email protected]> 2022-08-23 08:50 [PATCH v1] solaris: Use versioning scripts instead of -Bsymbolic Andres Freund <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v43 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v47 9/9] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v45 8/8] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v46 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v48 7/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v49 6/7] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]> 2026-03-17 19:22 [PATCH v44 10/10] Teach snapshot builder to skip transactions running REPACK (CONCURRENTLY). Antonin Houska <[email protected]>
This inbox is served by agora; see mirroring instructions for how to clone and mirror all data and code used for this inbox